My major theoretical goal is to develop integrated and compatible theories of speech perception, speech production and linguistic phonology. The research on speech perception investigates how listeners recover phonological segments (consonants and vowels) from a speech signal in which information for segments is thoroughly overlapped and context-sensitive. The research on production is aimed at discovering the coordinations among speech articulators that permit literal production of consonants and vowels in the vocal tract despite the temporal overlap among them that gives rise to context-sensitivity in the acoustic speech signal. The study of phonology is meant to discover ways of understanding linguistic “competence” so that its primitive components have only characteristics that are implementable in the vocal tract.
In a different line of research, I am investigating ways in which language performance reflects the changing base of knowledge shared between talker and listener. We are finding that at many levels of language (phonetic, lexical, syntactic) talkers reduce forms as shared knowledge grows. That is, talkers supply less information where less is needed; accordingly, when information is already foregrounded for the listener words are short durationally, lexically shorter words or phrases are used to refer and some optional syntactic markers may be omitted. We find that the patterning of reductions and elaborations itself marks episode units in a narration, and thus may provide useful information to a listener about discourse structure.